OMA-1 is a component of oocyte RNPs

C. elegans oocytes contain several different, and likely overlapping, types of RNPs. When sperm are absent, oocyte RNPs containing CAR-1, CGH-1, and several other RNA-binding proteins enlarge and become cortically localized, a process that is inhibited by the MSP signaling pathway (Jud et al. 2008; Noble et al. 2008). Similarly, GFP-tagged OMA-1, which is diffusely localized in the presence of sperm (Figure 2, E and F), is found in large cortical foci in the absence of sperm (Figure 2, G and H; Figure S1, D, F, N, and P). Large foci of OMA-1 are also found when MSP signaling is compromised in acy-4 mutants (Figure 2, I and J), but are absent from spe-9(ts) animals (Figure S1, G and H) that have fertilization-incompetent sperm able to stimulate meiotic maturation (Singson et al. 1998). In addition, OMA-1 foci in females are disrupted by RNAi-mediated knockdown of genes encoding the RNA-binding proteins PUF-5 and CAR-1 (Figure S1, Q and R; C. Spike, unpublished results), as described for CAR-1 and CGH-1-containing RNPs in female oocytes (Noble et al. 2008; Hubstenberger et al. 2013). These observations suggest that OMA-1 is a component of oocyte RNPs, at least in the absence of sperm.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

OMA-1 is a component of riboncleoprotein particles (RNPs). (A) OMA-1::S::TEV::GFP (asterisk) is depleted after incubation with matrix-coupled anti-GFP antibodies (compare lanes 1 and 3). OMA-1::S (double asterisk in A–C) is subsequently eluted from the affinity matrix by digestion with TEV protease. A total of 0.25% of each lysate and 1% of each TEV-eluted sample were loaded. Purifications and protease cleavage were monitored by western blotting using either anti-OMA-1 (shown), anti-S-tag, or anti-GFP antibodies. Here, and in subsequent panels, samples marked with a plus (+) sign were prepared from lysates containing OMA-1::S::TEV::GFP. Samples marked with a minus (−) sign are negative controls prepared from lysates lacking OMA-1::S::TEV::GFP. All purifications were performed in an oma-1(zu405te33) genetic background and were from fog-1(ts) females, unless otherwise specified. (B) Abundant proteins that copurify with OMA-1::S from fog-1 and spe-9 extracts were visualized by staining a polyacrylamide gel with SYPRO-Ruby (red boxes). Proteins in negative controls [minus (−) sign] are similar in size to human keratins (Moll et al. 2008), common contaminants of protein purifications. (C and D) Many proteins require RNA for their association with OMA-1. (C) Treatment with RNase A, prior to incubation with TEV protease (RNase elution, r⁺), causes proteins to elute from the immunoaffinity matrix. Proteins are not eluted by a mock RNase treatment (RNase elution, nr). Comparatively few proteins copurify with OMA-1::S after RNase A treatment (TEV elution, r⁺). Proteins were visualized using SYPRO-Ruby. (D) Western blots show that CGH-1 and CAR-1 copurify with OMA-1::S in an RNA-dependent manner. (E–J) OMA-1 reorganizes into large RNPs (arrowheads) when the sperm-dependent signal promoting meiotic maturation is absent or not transmitted to oocytes. Oocytes expressing the rescuing OMA-1::S::TEV::GFP fusion protein show a diffuse pattern of GFP localization (E and F), similar to spe-9(ts) animals raised at 25° (Figure S1). If sperm are absent, as in fog-1(ts) animals raised at 25°, OMA-1::S::TEV::GFP reorganizes into large foci (G and H). Similar foci form in the presence of sperm when the MSP-dependent signaling pathway active in sheath cells is disrupted, as in acy-4(ok1806) mutants (I and J). Explicit genotypes are specified in the legend to Figure S1. Bar, 20 μm.

OMA-1 and OMA-2 are oocyte-specific CCCH zinc finger RNA-binding proteins (Detwiler et al. 2001). The OMA proteins repress the translation of a few known target mRNAs in oocytes (Jadhav et al. 2008; Guven-Ozkan et al. 2010; Oldenbroek et al. 2013; Kaymak and Ryder 2013). After fertilization, OMA-1 and OMA-2 become phosphorylated and interfere with transcription by preventing TAF-4, a subunit of the TFIID RNA polymerase II transcriptional complex, from entering the nucleus (Guven-Ozkan et al. 2008). We isolated the mRNAs and proteins that associate with OMA-1 in oocytes using immunoaffinity purification (Figure S2). To avoid isolating OMA-1 complexes from embryos, protein extracts were made from sterile adults. These animals make normal oocytes but either lack sperm [fog-1(ts)] or have sperm that are unable to fertilize oocytes [spe-9(ts)] when grown at 25°. To facilitate purification, OMA-1 was tagged at the C terminus using a reversed version of the tag described by Cheeseman et al. (2004), which includes an S-tag, TEV protease cleavage site, and GFP. OMA-1::S::TEV::GFP is able to restore fertility to oma-1; oma-2 mutant animals, indicating that it functions properly in vivo. We conducted purifications in the oma-1(zu405te33) protein null mutant background (Detwiler et al. 2001) to avoid competition with the endogenous protein. Immunoaffinity purification was performed using anti-GFP antibodies followed by cleavage with TEV protease to elute OMA-1 complexes (Figure 2A; Figure S2) and reduce nonspecific background (Gerber et al. 2004). An S-tag/S-protein purification step would introduce RNase activity (Raines et al. 2000) and was not a part of our purification scheme; instead the S-tag was used to detect OMA-1::S after cleavage with TEV protease. Total protein stains indicate that numerous proteins copurify with OMA-1 (Figure 2B). Furthermore, the banding patterns appear similar in independent purifications (Figure 2, B and C; C. Spike, unpublished results), suggesting reproducible purification of the same collection of proteins. Furthermore, most proteins that copurify with OMA-1 are eluted from the affinity matrix by RNase treatment prior to cleavage with TEV protease (Figure 2C), indicating that their interactions with OMA-1 either require or are stabilized by RNA.

Identification of OMA-1-associated mRNAs

Because OMA-1/2 are essential for meiotic maturation, and function downstream of the MSP signaling pathway (Detwiler et al. 2001; Govindan et al. 2006, 2009; Kim et al. 2012), we hypothesized that OMA-1 might regulate, and therefore associate with, different mRNAs in the presence and absence of sperm-dependent MSP signaling. To test this hypothesis, we used Affymetrix microarrays to compare the mRNAs that copurify with OMA-1 in the presence and absence of sperm-dependent signaling [i.e., spe-9(ts) and fog-1(ts) strains, respectively]. Contrary to our expectation, essentially the same mRNAs were identified as significantly enriched in OMA-1 purifications relative to total lysate mRNA (input mRNA) in three biological replicates for both sets of OMA-1 purifications (Figure 3A; ∼1079 mRNAs corresponding to 1290 probe sets enriched at least twofold). Furthermore, when we directly compared the OMA-1 purifications from spe-9(ts) and fog-1(ts) animals, only two mRNAs were significantly enriched only in purifications from either strain (File S1), and neither mRNA was significantly enriched in OMA-1 purifications compared to input mRNA. Since several of the enriched mRNAs are in vivo targets of OMA-dependent translational repression (see below), these observations suggest that OMA-1 stably associates with the same mRNA targets in the presence and absence of sperm-dependent MSP signaling. We consider mRNAs identified as OMA-1-associated in both sets of purifications to be the best candidates for targets of OMA-dependent regulation in vivo (Figure 3A; File S1). After removing duplicates, this list of ∼1079 genes includes all previously identified mRNA targets of OMA-dependent translational repression in oocytes: zif-1, mom-2, nos-2, and glp-1 (Jadhav et al. 2008; Guven-Ozkan et al. 2010; Oldenbroek et al. 2013; Kaymak and Ryder 2013). mei-1 mRNA was not identified (File S1), likely because this mRNA is a target of OMA-1-mediated repression in embryos, but not oocytes (Li et al. 2009), which was the basis for our purification.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Identification and analysis of mRNAs that copurify with OMA-1. (A) GeneChip microarrays identified OMA-1-associated mRNAs (1290 probe sets) that are enriched at least twofold in OMA-1 purifications relative to input samples [P(corr) ≤ 0.05]. Similar sets of mRNAs were identified in OMA-1 purifications from fog-1(ts) (red circle) and spe-9(ts) (blue circle) lysates, including known targets of OMA-dependent translational repression (zif-1, mom-2, nos-2, and glp-1). Note that most mRNAs that are significantly enriched in a single experiment are enriched less than twofold, or are enriched but fail data-quality metrics, in the other experiment (File S1). (B) The relative abundances of candidate target mRNAs in a representative OMA-1 purification were determined by high-throughput sequencing. All candidate mRNA targets with FPKM values >1 are shown; colors indicate FPKM values (red > 569, yellow = 127, blue < 18.5). Genes whose 3′-UTRs have been tested for their ability to confer OMA-dependent translational repression in oocytes are shown; those with positive results are in black. Two targets of OMA-dependent translational repression have multiple mRNA isoforms with FPKM values >1 (asterisks); only the most abundant isoform is indicated. (C) A significant percentage of OMA-1 target genes are highly expressed during oogenesis (oogenesis-associated) or identified as germline-intrinsic (P = 1.4 × 10^–21). Enrichment of both gene classes is most dramatic among OMA-1 targets of high abundance in the immunopurified samples (P = 2.3 × 10^–95). The FPKM value of the most abundant transcript of each gene was used to estimate target abundance. Significance was determined using a hypergeometric probability test. (D) OMA-1 target mRNAs tend to have more OMA-1-binding motifs in their 3′-UTRs than other mRNAs with similarly sized 3′-UTRs. The numbers of UA[A/U] motifs found in a genome-wide collection of C. elegans UTRs (Jan et al. 2011) are plotted relative to 3′-UTR length (black circles). All 3′-UTRs <950 nucleotides were plotted (98.2% of identified UTRs). Red- and yellow-filled circles correspond to likely and known targets of OMA-dependent translational repression, respectively (n = 120). Linear regression of these datasets generated lines with similar slopes but significantly different y-intercepts (P < 0.0001), suggesting that there are additional UA[A/U] motifs in the 3′-UTRs of likely OMA-1 target mRNAs.

We noticed that zif-1, mom-2, nos-2, and glp-1 appear to be more abundant in OMA-1 purifications than many other OMA-1-associated mRNAs (Figure S3). Because the relative levels of different mRNAs in the same sample cannot be reliably measured using microarrays (Draghici et al. 2006), we used Illumina high-throughput sequencing to identify and quantify the mRNAs present in a representative OMA-1 purification. The mRNA sample selected was also analyzed in the microarray experiments and found to be similar to other OMA-1 purifications (Spearman rank-order correlations r_s ≥ 0.967). Only 42% of the mapped reads from this sample correspond to rRNA, which represents a significant depletion considering that no rRNA depletion or mRNA enrichment steps other than OMA-1 purification were performed (see Materials and Methods), and total RNA is >80% rRNA. We eliminated rRNA sequences prior to determining transcript abundance in the immunopurified samples, which used the FPKM measurement developed by Trapnell et al. (2010). Indeed, zif-1, mom-2, nos-2, and glp-1 are more abundant than most other OMA-1-associated mRNAs (Figure 3B) and among the 200 most abundant mRNAs in the immunopurified sample analyzed (File S1). We verified that this analysis provides an accurate estimate of the relative levels of OMA-1-associated mRNAs across all samples by comparing to quantitative RT-PCR data. In the sequenced sample, zif-1 has a 5.1-fold higher FPKM value than nos-2, and zif-1 was consistently 4–7-fold more abundant than nos-2 by qRT-PCR in independent OMA-1 purifications (Table S2).

We next assessed whether OMA-1-associated mRNAs are present in oocytes; such mRNAs could be in vivo components of OMA-1 RNPs. Genes that are highly expressed during oogenesis or identified as germline intrinsic by Reinke et al. (2004) are likely expressed in oocytes and are significantly enriched among genes encoding OMA-1-associated mRNAs (Figure 3C; P = 1.4 × 10⁻²¹). Enrichment of these gene categories was even more significant among genes encoding OMA-1-associated mRNAs of high abundance in the immunopurifications (Figure 3C; P = 2.3 × 10⁻⁹⁵). OMA-associated mRNAs that are abundant in OMA-1 purifications are therefore the most likely to be present in oocytes and in vivo targets of OMA-dependent translational repression. We examined the biological processes associated with these high-confidence mRNA targets of OMA-1 using DAVID tools (http://david.abcc.ncifcrf.gov) to identify enriched Gene Ontology (GO) terms (Huang et al. 2009a,b). Enriched GO terms and GO term clusters related to reproduction, embryonic development, and germline sex determination were identified (see File S1 for complete lists), consistent with identified functions of OMA-1/2 in vivo (see Discussion).

OMA-1/2 regulate the translation of OMA-1-associated mRNAs

Sequences in the 3′-UTRs of mRNAs are important for regulating protein accumulation in the oogenic germ line (Merritt et al. 2008) and mediate the OMA-dependent translational repression of zif-1, mom-2, nos-2, and glp-1 in oocytes (Jadhav et al. 2008; Guven-Ozkan et al. 2010; Oldenbroek et al. 2013; Kaymak and Ryder 2013). To examine the regulation of OMA-1-associated mRNAs in vivo, we generated animals expressing reporter transgenes containing the 3′-UTRs of eight different candidate targets (Figure 3B) and fbf-2, a germline-expressed mRNA that is not OMA-1-associated (File S1). Each transgene uses the pie-1 promoter to express a GFP::histone 2B (H2B)-coding mRNA in the germ line, but has a distinct 3′-UTR sequence. Based on genome-wide expression profiles, all of the OMA-1-associated mRNAs we tested for 3′-UTR-dependent regulation are plausibly present in oocytes and maternally provided to embryos (Baugh et al. 2003; Reinke et al. 2004; Gerstein et al. 2010). As expected, GFP expression from the fbf-2 reporter transgene was unaffected by oma-1/2(RNAi) (Figure 4, G and H). However, six of the eight transgenes containing the 3′-UTRs of OMA-1-associated mRNAs had higher levels of GFP expression in oma-1/2-depleted oocytes compared to control oocytes (Figure 4, A–F; Figure S4, A–F). Increased GFP expression requires the simultaneous depletion of oma-1 and oma-2; GFP levels appear unchanged after oma-1(RNAi) if oma-2 is undepleted and the converse is also true (D. Coetzee and C. Spike, unpublished results). Interestingly, reporter constructs containing the cdc-25.3, rnf-5, and rnp-1 3′-UTRs were more strongly derepressed after oma-1/2(RNAi) than the other reporters (fce-1, pqn-70, and gap-2 3′-UTRs). We quantified these changes after crossing 3′-UTR transgenes into oma-1(zu405te33) or oma-2(te51) loss-of-function mutants, because single-gene RNAi tends to be more effective (Gönczy et al. 2000). Oocytes expressing reporter constructs containing the cdc-25.3, rnf-5, and rnp-1 3′-UTRs had 5- to 18-fold more GFP in Oma oocytes compared to controls, and these increases were highly significant (Figure 4, I–K). Oma oocytes expressing reporter constructs containing the fce-1 and pqn-70 3′-UTRs also had significant increases in GFP expression, but were more modestly affected with only two- to threefold more GFP compared to controls (Figure S4, K and L). Distinct spatial–temporal patterns of translation were noted for the different 3′-UTR reporter constructs regulated by OMA-1 and OMA-2. Reporters containing the cdc-25.3 and rnp-1 3′-UTRs are repressed in oocytes, but expressed in fairly young embryos (Figure S5). Similar patterns of regulation have been noted for other targets of OMA-dependent translational repression in oocytes (Evans et al. 1994; Subramaniam and Seydoux 1999; Guven-Ozkan et al. 2010, Oldenbroek et al. 2012, 2013), suggesting that the OMA proteins function—at least in part—to repress maternally provided mRNAs that are translated during embryogenesis.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

The 3′-UTRs of mRNAs that copurify with OMA-1 can mediate OMA-dependent translational repression in oocytes. GFP::histone 2B (H2B) expression from a 3′-UTR reporter transgene from either cdc-25.3 (A, B, and I), rnf-5 (C, D, and J), or rnp-1 (E, F, and K) is strongly increased in oocytes after reducing oma-1/2 function. GFP::H2B expression from a reporter transgene containing the fbf-2 3′-UTR (G and H), which is not OMA-1-associated, is unaffected. The images shown here (and in Figure S4, Figure S5, Figure S6, and Figure S7) often include germline pachytene nuclei, which are smaller than and dorsal to the oocyte nuclei (arrowheads). oma-1/2 function was compromised using oma-1/2(RNAi) (B, D, F, and H), by combining RNAi with a loss-of-function mutation in oma-1 or oma-2 (I and K) or by crossing the transgene into oma-1(te33zu405); oma-2(te51) double mutants (J). Background-corrected nuclear GFP intensity (in arbitrary fluorescence units, plotted on the y-axis) was measured for the three oocytes closest to the spermatheca (x-axis, oocytes –1 to –3) and is significantly increased in oma-1/2 oocytes in every position compared to controls (two asterisks, P < 0.001 using a Mann–Whitney U-test). Box plots represent the data from 10 (I and K) or 28–29 (J) oocytes in each position. Box plot whiskers indicate the minimum and maximum intensity values. Bar, 20 μm.

Oma oocytes do not undergo meiotic maturation and remain in the gonad indefinitely. We examined 3′-UTR reporters regulated by OMA-1/2 in fog-2(oz40) female and gsa-1(RNAi)-treated oocytes, which mature infrequently because MSP-dependent signaling is absent (fog-2) or has been compromised [gsa-1(RNAi); Govindan et al. 2006, 2009]. The amount of GFP::H2B expressed in these oocytes was not dramatically different from the wild type (Figure S6), suggesting that reduced rates of oocyte maturation per se do not cause or substantially contribute to increased GFP expression in Oma oocytes. Furthermore, these results suggest that the translational repression of OMA targets occurs even in the absence of MSP-dependent signaling. Indeed, oma-1/2 depletion in fog-2 females increases GFP expression from the OMA-target 3′-UTR transgenes tested (cdc-25.3, rnp-1, and rnf-5; Figure S7). Thus, OMA-1 and OMA-2 are required for repression in the absence of MSP-dependent signaling, which is consistent with the result that substantially the same mRNAs copurify with OMA-1 irrespective of the presence of the MSP signal. Collectively, these results are consistent with the idea that OMA-1 and OMA-2 repress the translation of many OMA-1-associated mRNAs in oocytes, independent of the MSP signaling pathway and suggest that OMA-dependent translational repression can be relatively weak or strong, depending on the 3′-UTR of the mRNA target. Further these results indicate that the change in OMA-1 RNP localization in the absence of MSP-dependent signaling is independent of its translational repression activity.

OMA-1-binding motifs in the 3′-UTRs of OMA-1-associated mRNAs

OMA-1 binds with high affinity to RNAs that contain multiple copies of a short UA[A/U] consensus sequence in vitro (Kaymak and Ryder 2013). However, OMA-1-binding motifs are extremely common in C. elegans 3′-UTRs (>99% have at least one OMA-1-binding motif; Figure 3D), which tend to be AU-rich (Mangone et al. 2010; Jan et al. 2011). Thus, we examined whether OMA-1-binding motifs are unusually prevalent in the 3′-UTRs of OMA-1-associated mRNAs. For this analysis, we used 120 OMA-1-associated mRNAs that are relatively abundant in OMA-1 purifications or established targets of OMA-1-dependent translational repression in oocytes (Figure 3B) and chose the 3′-UTR identified by Jan et al. (2011) that most closely matches our RNAseq data (File S1). Such OMA-1-associated 3′-UTRs tend to be longer than typical C. elegans 3′-UTRs (median length of 205 instead of 130 nucleotides), consistent with the idea that they contain important regulatory sequences. Although there was a poor correlation between the absolute number of OMA-1-binding motifs in each 3′-UTR and mRNA abundance after OMA-1 purification (R² = 0.09), OMA-1-binding motifs tend to be slightly enriched in OMA-1-associated 3′-UTRs relative to similarly sized C. elegans 3′-UTRs (Figure 3D), suggesting that a subset of OMA-1-binding motifs might be important for OMA-1 binding. We examined whether a longer consensus sequence containing UA[A/U] motifs was present in this collection of OMA-1-associated 3′-UTRs, but were unable to identify one (using MEME; Bailey et al. 2009). Together, these results suggest that at least some OMA-1-binding motifs may be important for OMA-1 binding, but also imply that additional determinants of OMA-1-binding specificity likely exist, as postulated by Kaymak and Ryder (2013). Our finding that the set of OMA-1-associated proteins includes many RNA-binding proteins (see below) is consistent with this view.

Involvement of OMA-1 target mRNAs in oogenesis

The set of OMA-associated mRNAs and proteins (see below) will provide insight into the role of OMA-1 and OMA-2 in the regulation of oocyte meiotic maturation and the oocyte-to-embryo transition, a point illustrated in the companion article (Spike et al. 2014). We initially focused on cdc-25.3 and rnp-1 because their 3′-UTRs appear to mediate strong translational repression by OMA-1 and OMA-2 and both genes are highly conserved (Shaye and Greenwald 2011). cdc-25.3 encodes one of four members of the Cdc25 family of phosphatases that activate meiotic maturation by removing inhibitory phosphorylations of CDK at residues Thr14 and Tyr15 (Kumagai and Dunphy 1991), catalyzed by the Wee1 or Myt1 kinases (Kornbluth et al. 1994; Mueller et al. 1995). The identification of a CDK activator as a target of OMA-1 and OMA-2-mediated translational repression was counterintuitive because oocytes in oma-1; oma-2 double mutants fail to undergo meiotic maturation. Nonetheless, we tested whether the regulation or function of cdc-25.3 might be critical for oogenesis, a possibility that later proved correct, though not in the ways we originally imagined (Spike et al. 2014). We first examined the phenotype of an existing deletion in cdc-25.3 to determine whether its function is important for oocyte or embryo development. cdc-25.3(ok358) animals are viable and fertile, but exhibit partially penetrant larval arrest, as described in our companion article (Spike et al. 2014). Additional analysis suggests that cdc-25.3 functions during embryogenesis but is redundant with cdc-25.2, a different cdc25-family gene important during oogenesis (Kim et al. 2010a); weak cdc-25.2(RNAi) causes highly penetrant embryonic lethality in cdc-25.3(ok358) animals (100%; n = 298), but not in the wild type (3%; n = 349). Potentially, overexpression of CDC-25.3 in oma-1; oma-2 double mutants might interfere with CDK-1 activation. We tested this possibility genetically by constructing cdc-25.3(ok358); oma-1(zu405te33); oma-2(te51) triple mutants, which were found to exhibit the Oma phenotype and were completely sterile.

Although OMA-1 and OMA-2 negatively regulate the translation of 3′-UTR reporter constructs, their target mRNAs might be destabilized in their absence. We tested this possibility using fluorescent in situ hybridization (FISH) to examine the spatial–temporal pattern of endogenous cdc-25.3 and rnp-1 mRNA accumulation in the germ line. Both OMA-1-associated mRNAs are present in oocytes (Figure 5, A and F). cdc-25.3 mRNA is first detected in late pachytene (Figure 5A), while rnp-1 mRNA is detected throughout the germ line (Figure 5, F and I; D. Coetzee, unpublished results). cdc-25.3 and rnp-1 mRNAs are also clearly present in oma-1; oma-2 mutant germ lines and oocytes (Figure 5, C, D, G, and J). Finally, cdc-25.3 mRNA was not detected in the oocytes of the cdc-25.3(ok358) deletion mutant (Figure 5B), confirming the specificity of this probe and procedure. A similar experiment was not feasible for rnp-1 due to germline abnormalities in the deletion mutant (see below), but this probe is predicted to be highly specific (D. Coetzee, unpublished results). We conclude that OMA-1-associated mRNAs are not degraded in the absence of OMA-1 and OMA-2. We cannot exclude the possibility that transcript abundances are modestly altered, however, because older Oma oocytes are somewhat difficult to image after FISH and the localization patterns of both mRNAs change when meiotic maturation is inhibited. rnp-1 mRNAs coalesce, or aggregate, into higher order structures in oma-1; oma-2 mutant and fog-2 female germ lines, both in the rachis (Figure 5, J and K) and in oocytes (Figure 5, G and H), similar to what has been described for total RNA and a few specific mRNAs in female germ lines (Schisa et al. 2001; Jud et al. 2008; Noble et al. 2008). Some aggregation of the cdc-25.3 mRNA was also observed (Figure 5, C and E), but its localization more closely resembles the wild type, particularly in oma-1; oma-2 mutant oocytes (Figure 5D). The highly similar fbf-1 and fbf-2 mRNAs (Zhang et al. 1997), which are not OMA-1-associated (File S1), also aggregate in oma-1; oma-2 mutant and fog-2 female germ lines (Figure 5, L–N; D. Coetzee, unpublished results), although not as dramatically as rnp-1 mRNA.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

cdc-25.3 and rnp-1 mRNAs are present in wild-type and oma-1/2 oocytes. Fluorescent in situ hybridization (FISH) was used to examine cdc-25.3 (A–E), rnp-1 (F–K) and fbf-1/2 (L–N) mRNA expression and localization in wild-type (A, F, I, and L), oma-1(zu405te33); oma-2(te51) (C, D, G, and J), and fog-2(oz40) (E, H, K, M, and N) germ lines. mRNAs are visualized as small yellow puncta; DAPI-stained DNA is magenta. cdc-25.3 mRNAs are absent from the distal germ line, present during late pachytene, diplotene, and diakinesis (A), and absent from cdc-25.3(ok358) mutants (B). cdc-25.3 mRNA expression in oma-1/2 and fog-2 germ cells is similar to wild type, although there is some aggregation of individual mRNAs in late pachytene in oma-1/2 animals (C) and in older fog-2 oocytes (E) and possibly somewhat fewer puncta in older oma-1/2 oocytes (not shown). rnp-1 and fbf mRNAs were present throughout the germ line, including pachytene-stage germ cells (I) and oocytes (F and L). Both mRNAs aggregate in the rachis (J, K, and M) and in oocytes (G, H, and N) of oma-1/2 and fog-2 animals compared to the wild type. Aggregation of the rnp-1 mRNA was more dramatic than the aggregation of either cdc-25.3 or fbf mRNAs and was most distinct in fog-2 animals. Bar, 20 μm.

While cdc-25.3 is dispensable for oogenesis, rnp-1 appears to be required. Adult rnp-1(ok1549) hermaphrodites are completely sterile with small germ lines that fail to transition from spermatogenesis to oogenesis (Figure 6B). Genetic analysis has identified many of the key genes that control sex determination in C. elegans (reviewed by Ellis and Schedl 2007; Kimble and Crittenden 2007). fog-3 is one of the most downstream genes in the germline sex-determination pathway and is required for spermatogenesis (Chen et al. 2000). Furthermore, the fog-3 gene is epistatic to rnp-1 with respect to germline sex determination. fog-3; rnp-1 double mutants fail to make sperm, or express MSP, but some animals make tiny underdeveloped oocytes (6 of 13; Figure 6C). Thus, rnp-1 functions upstream of fog-3 in the germline sex determination pathway and promotes both oogenesis and normal oocyte development. Since rnp-1 mRNA is present throughout the germ line (Figure 5, F and I; D. Coetzee, unpublished results), but the OMA proteins are first expressed in late pachytene (Detwiler et al. 2001; Figure 1), RNP-1 is likely expressed earlier in germline development.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6

rnp-1 promotes the switch from spermatogenesis to oogenesis and is required for normal oocyte development. (A) Wild-type animals make both sperm (red, anti-MSP) and oocytes (green, anti-RME-2). (B) rnp-1(ok1549) animals have small germ lines with sperm (red) but no oocytes. Nuclei transitioning from mitosis to meiosis (tz) and in the pachytene stage of meiotic prophase are indicated in B and C. (C) fog-3(q470) unc-13(e1091); rnp-1(ok1549) animals lack sperm but can make tiny underdeveloped oocytes (green). Six of 13 germ lines examined expressed RME-2 and none expressed MSP. Bar, 20 μm.

Identification of OMA-1-associated proteins

We next investigated the proteins that copurify with OMA-1. OMA-1 RNPs were isolated in the presence and absence of MSP-dependent signaling and copurifying proteins were identified using mass spectrometry. OMA-1 was well covered (≥59%) and identified by a large number of peptides in both purifications (≥60 peptides; File S2). Most of the other proteins identified were also present in both purifications (Figure S8), including the abundantly represented proteins CGH-1 and CAR-1 (File S2). Both RNA-binding proteins are components of germline RNPs (Boag et al. 2005) and copurify with OMA-1 in the presence and absence of sperm (Figure 2D, File S2). Interestingly, western blots consistently detect CGH-1 and CAR-1 in OMA-1 purifications from fog-1(ts) extracts (n ≥ 3), but do not detect either protein in purifications from spe-9(ts) extracts (n = 3), suggesting that CGH-1 and CAR-1 are more abundant in purifications from fog-1(ts) females, though mass spectrometry clearly shows they are present in purifications from spe-9(ts) extracts (File S2). Finally, the interaction of CGH-1 and CAR-1 with OMA-1 is RNA dependent (Figure 2D), as observed for many other proteins that copurify with OMA-1 RNPs (Figure 2C; File S2).

Mass spectrometry is extremely sensitive, and >250 different proteins were identified by at least two peptides in both OMA-1 RNP purifications (Figure S8, File S2). Many of these proteins, like CGH-1 and CAR-1, have RNA-related functions and could be important components of OMA-1 RNPs in vivo. Other copurifying proteins likely represent abundant contaminants (e.g., UNC-54/myosin and VIT-1/vitellogenin; see Materials and Methods). We focused on the subset of proteins that copurify with OMA-1 from fog-1(ts) females after RNase treatment based on the expectation that close associations with OMA-1 (be they direct or indirect) might be at least partially resistant to RNase treatment. Many proteins, including some with RNA-related functions (e.g., CGH-1, EDC-3, and CEY-4), were depleted from OMA-1 RNPs by RNase treatment, leaving a much smaller pool of candidates (133 different proteins; Figure S8). Importantly, the eIF4E-binding protein IFET-1 continued to copurify with OMA-1 in the presence of RNase A (File S2). Prior work established that IFET-1 interacts with OMA-1 in vitro and represses the translation of OMA target 3′-UTR reporters in vivo (Li et al. 2009; Guven-Ozkan et al. 2010; Oldenbroek et al. 2013). Next, proteins identified in negative controls or as abundant contaminants were excluded from consideration, leaving a smaller list of OMA-1-associated proteins (51 different proteins; Figure S8). This step eliminated CAR-1, but again retained IFET-1 (File S2). It is difficult to eliminate all contaminants identified by mass spectrometry using a limited number of negative controls (Mellacheruvu et al. 2013), so we examined the biological functions of the remaining proteins in more detail (Table 1, File S2).

Many OMA-1-associated proteins have RNA-related functions (27 of 51; 53%), including mRNA translation, cytoplasmic polyadenylation and deadenylation (Table 1, File S2), and most of these proteins function or are found in C. elegans oocytes (see below; reviewed in Nousch and Eckmann 2013). OMA-2 is one of these proteins, suggesting that OMA-1 and OMA-2 are closely associated in oocytes. OMA-1-associated proteins such as IFET-1, PUF-5, MEX-1, MEX-3, POS-1, and SPN-4 (Table 1) are present in oocytes and regulate the translation of OMA-1-associated mRNAs, either in oocytes (IFET-1 and PUF-5) or in early embryos (Ogura et al. 2003; Lublin and Evans 2007; Jadhav et al. 2008; Pagano et al. 2009; Guven-Ozkan et al. 2010; Oldenbroek et al. 2012, 2013). IFET-1 interacts with multiple eIF4E isoforms (Li et al. 2009), including the OMA-1-associated isoform IFE-3 (14% coverage, Table 1), which is one of the germline-enriched eIF4E proteins in C. elegans (Amiri et al. 2001). eIF4E interacts with eIF4G/IFG-1 (10% coverage, Table 1), which is important for oocyte development in C. elegans (Contreras et al. 2008); these proteins form complexes that both inhibit and activate translation (Hentze 1997; Rajyaguru et al. 2012). OMA-1-associated proteins important for cytoplasmic polyadenylation include GLD-2 (16% coverage; Table 1) and GLD-3 (21% coverage; Table 1), which are present in oocytes and required for normal oocyte development (Kadyk and Kimble 1998; Kim et al. 2010b). GLD-2 is the catalytic poly(A)polymerase subunit (Wang et al. 2002), and many of the GLD-2-associated transcripts identified by Kim et al. (2010b) are also OMA-1-associated (52%; 284/544). Finally, OMA-1-associated proteins important for deadenylation include CCF-1 and LET-711/NTL-1, the two Ccr4–Not complex subunits shown to be crucial for fertility and normal oocyte development (Nousch et al. 2013). Interestingly, components of the Ccr4–Not complex were found to interact with the mammalian TIS11 zinc-finger protein tristetraprolin (reviewed by Brooks and Blackshear 2013). Thus, many of the proteins identified as OMA-1-associated are important components of oocyte or germline RNPs and possibly relevant to OMA-1 function in vivo.

OMA-1-associated proteins interact with OMA-1

Yeast two-hybrid assays suggest that OMA-1 physically interacts with several of the proteins identified by mass spectrometry (Table 1; Figure 7, A and B). Importantly, these interactions were identified independent from, and blind to, the mass spectrometry results and represent an orthogonal dataset. Instead, candidate proteins were identified in a yeast two-hybrid screen (PQN-59) or tested for a physical interaction with OMA-1 based on in vivo biological function (SPN-4, MEX-3, GLD-1, and OMA-2). The interactions of SPN-4, MEX-3, and GLD-1 with OMA-1 map to a region containing the two OMA-1 zinc fingers (amino acids 111–188; Y. Nishi and R. Lin, unpublished results), and are abolished by the E141K zinc finger mutation (Figure 7A). However, the amino terminus of OMA-1 and OMA-2 mediates the yeast two-hybrid interaction between these two proteins (amino acids 1–117 and 1–111, respectively; Figure 7B). Each interaction is of weak-to-moderate strength, but stronger than negative controls and comparable to the positive control TAF-4 (Guven-Ozkan et al. 2008; Figure 7, A and B). Combined with previous data indicating that IFET-1 interacts with OMA-1 (Li et al. 2009), these results indicate that OMA-1 physically interacts with some of the proteins that regulate the translation of OMA target mRNAs. Notably, our OMA-1 RNP purifications identified all the previously known and newly identified OMA-1-interacting proteins save for three exceptions: TAF-4, C27B7.2, and DH11.5. The TAF-4–OMA-1/2 interaction requires MBK-2 phosphorylation of OMA-1/2, which occurs in the embryo (Guven-Ozkan et al. 2008), and thus we might not expect its representation in purifications from sterile adult hermaphrodites and females. Less is known about C27B7.2 and DH11.5, including their expression patterns, so their absence in our purifications is not easily interpreted.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7

Several proteins that copurify with OMA-1 interact with OMA-1 in yeast two-hybrid assays. (A) MEX-3, SPN-4, GLD-1, and PQN-59 interact with OMA-1 in GAL4-based yeast two-hybrid assays on 10 mM 3-amino-1, 2,4-triazole (3AT). Bait vectors are underlined; ZYG-11 bait and prey vectors are negative controls. TAF-4, C27B7.2, and DH11.5 were identified in a GAL4-based yeast two-hybrid screen for proteins that interact with the OMA-1 N terminus (Guven-Ozkan et al. 2008; R. Lin, unpublished results) and are positive controls. PQN-59 was independently identified as an OMA-1-interacting protein using an SRS-based yeast two-hybrid screen (R. Lin, unpublished results). MEX-3, SPN-4, and GLD-1 were tested as candidate OMA-1-interacting proteins based on their expression patterns and biological functions. OMA-1 E141K is a point mutation that affects the first OMA-1 zinc finger; this residue is critical for OMA-1 function in vivo (Detwiler et al. 2001). MEX-3, SPN-4, and GLD-1 all interact with a region of OMA-1 containing the CCCH zinc fingers in similar assays (Y. Nishi and R. Lin, unpublished results). (B) OMA-1 and OMA-2 interact in a yeast two-hybrid assay. The interaction of the N-terminal domains of OMA-1 and OMA-2 is weaker than the interaction of full-length OMA-1 with the OMA-2 N terminus. Full-length OMA-1 and OMA-2 also interact, but less reproducibly (T. Guven-Ozkan and R. Lin, unpublished results). Yeast strains used were AH109 (A) and Mav203 (B).

puf-5(RNAi) exhibits synthetic lethality with oma-2

The set of proteins identified in our immunopurifications comprise most or all independently defined OMA-1-interacting proteins. Thus, we expect that this set will prove informative for understanding the biological functions of OMA-1/2. To begin to assess the relevance of this set, we conducted a screen for suppressors and enhancers of oma-1 and oma-2. We used gene-specific RNAi to reduce the function of most OMA-1-associated proteins (File S2). These experiments focused on OMA-1-associated proteins with RNA-related functions, but included many other proteins present in OMA-1 purifications, including CGH-1 and CAR-1. RNAi was performed in a variety of genetic backgrounds, including the wild type, oma-1(zu405te33); oma-2(te51) double mutants, and oma-1(zu405te33) and oma-2(te51) single mutants. No suppressors of the oma-1; oma-2 oocyte meiotic maturation defect were identified. This analysis uncovered a synthetic lethal interaction between puf-5(RNAi) and oma-2(te51). puf-5(RNAi) causes low-penetrance embryonic lethality in the wild type and oma-1(zu405te33) mutants (9%; n ≥ 172) but high-penetrance embryonic lethality in oma-2(te51) mutants (97%; n = 223). oma-2(te51) mutants do not exhibit appreciable embryonic lethality in control RNAi experiments (2%; n = 85). This result suggests that PUF-5 and OMA-2 function redundantly and, somewhat unexpectedly, that OMA-1 and OMA-2 are not completely redundant either during oogenesis or early embryogenesis. Furthermore, it confirms that PUF-5 is likely relevant to OMA function in vivo, as surmised from the observation that each of these proteins represses the translation of glp-1 mRNA in oocytes (Lublin and Evans 2007; Kaymak and Ryder 2013).

LIN-41 represses the translation of OMA-1/2 targets in oocytes

In a second approach for identifying proteins relevant to OMA-1/2 function, we conducted an RNAi screen for regulators of 3′-UTR-dependent OMA-1/2-mediated translational repression. Animals expressing the zif-1 3′-UTR reporter were used in our initial RNAi screen because this reporter is strongly repressed in wild-type oocytes (Figure 8, A, I, and M; Guven-Ozkan et al. 2010). Interestingly, IFET-1 and LIN-41 were the only OMA-1-associated proteins we identified that strongly repress zif-1 translation in oocytes. IFET-1 was previously shown to repress OMA-1/2 targets in oocytes (Li et al. 2009; Guven-Ozkan et al. 2010; Oldenbroek et al. 2013). LIN-41 is a conserved member of the TRIM-NHL family of proteins and has been proposed to repress mRNA translation in C. elegans as well as other organisms (Slack et al. 2000; Loedige et al. 2013; Worringer et al. 2014). GFP expression from the zif-1 3′-UTR reporter was increased after lin-41(RNAi) near the loop region (Figure 8, A and B); these germ cells are young abnormal oocytes and are described in detail in our companion article (Spike et al. 2014). We examined other 3′-UTR reporters strongly repressed by OMA-1/2 to see if they are also repressed by LIN-41 in oocytes. Indeed, GFP expression from the cdc-25.3 and rnp-1 3′-UTR reporter contructs is dramatically increased after lin-41(RNAi) (Figure 8, C–F). lin-41(RNAi) may modestly increase GFP expression from the rnf-5 3′-UTR construct, but this transgene was much less strongly regulated (C. Spike, unpublished results). We also examined GFP expression from a subset of OMA target 3′-UTR constructs in the oocytes of strong loss-of-function alleles of lin-41, including the temperature-sensitive sterile lin-41(tn1487ts) allele (Spike et al. 2014). As expected, GFP expression from the cdc-25.3 and zif-1 3′-UTR constructs is increased in lin-41(tn1487ts) oocytes at the restrictive temperature (25°) relative to controls. GFP expression from the cdc-25.3 3′-UTR construct was robust and pervasive (Figure 8, K and L), while expression from the zif-1 3′-UTR construct was relatively faint and predominantly visible in young oocytes near the loop (Figure 8, M and N). A similar pattern was observed when the zif-1 3′-UTR construct was crossed into the strong loss-of-function lin-41(n2914) allele, suggesting that, in these mutants, the zif-1 3′-UTR construct is translated during early oogenesis. We confirmed that GFP is expressed from the zif-1 3′-UTR construct in lin-41(n2914) oocytes by examining young animals that are just beginning to produce oocytes. In such animals, GFP expression is observed in most, if not all, oocytes and appears to correlate temporally with oocyte formation (Figure 8, O and P).

• Download figure
• Open in new tab
• Download powerpoint

Figure 8

LIN-41 represses the translation of OMA targets in oocytes. (A–H) lin-41(RNAi) (B, D, F, and H) strongly enhances GFP::H2B expression from reporter transgenes containing the zif-1 (A and B), cdc-25.3 (C and D), or rnp-1 (E and F) 3′-UTRs in oocytes. There was no apparent increase in GFP::H2B expression from the reporter transgene containing the control fbf-2 3′-UTR (G and H). (I and J) GFP::H2B expression from the zif-1 reporter transgene can be seen in the oocytes of fertile lin-41(ma104) animals (67%; n = 54), but not wild-type controls (0%; n > 36). (K–N) GFP::H2B expression from the cdc-25.3 (K and L) and zif-1 (M and N) reporter transgenes is increased in the oocytes of lin-41(tn1487ts) mutants raised at 25° relative to controls. (O and P) Expression from the zif-1 reporter transgene (green) begins as lin-41(n2914) germ cells start to develop into oocytes (O); the most proximal oocyte is GFP positive (arrowhead). The germ line of this lin-41(n2914) animal was imaged just prior to vulval eversion (P), which would normally occur around the time of the L4-to-adult molt (Sharma-Kishore et al. 1999). Identical exposure times were used to collect the paired images with the exception that the control in A was 40% overexposed compared to B. Bar, 20 μm.

Strong losses of lin-41 function [as in lin-41(n2914), lin-41(tn1487ts), or lin-41(RNAi)] cause severe defects in oocyte growth and meiotic progression (Spike et al. 2014). However, GFP expression from OMA-1/2 target 3′-UTR constructs does not appear to be a secondary consequence of these abnormalities. GFP expression from the fbf-2 3′-UTR construct is not increased or altered after lin-41(RNAi) (Figure 8, G and H), indicating that lin-41 oocytes are still capable of repressing the translation of some mRNAs. Furthermore, GFP is visibly expressed from the zif-1 3′-UTR construct in lin-41(ma104) oocytes, but not wild-type oocytes (Figure 8, I and J). lin-41(ma104) is a hypomorphic allele that is viable and fertile (Slack et al. 2000). The oocytes produced by homozygous lin-41(ma104) animals tend to be small but appear relatively normal and do not exhibit defects in meiotic prophase progression (Figure 8J, C. Spike, unpublished results). After outcrossing, homozygous lin-41(ma104) animals exhibited only low levels of embryonic lethality (2%; n = 1765) and a brood size of 181 progeny (n = 12) at 20°. This is ∼56% of the wild-type brood size (∼320 progeny), consistent with the observation that oocyte development is only modestly impaired in lin-41(ma104) animals. Together, these results indicate that LIN-41 represses the translation of several OMA targets in oocytes and is likely relevant to OMA function in vivo as we show in the accompanying article (Spike et al. 2014).